Study On The Damage Identification Of Offshore Wind Turbine Support Structure Under Operational Condition

The offshore wind turbine support structure is located in a complex and harsh environment and is liable to be damaged.In order to prevent the catastrophic accidents such as collapse,it is necessary to carry out the study on the damage identification of offshore wind turbine support structures.Under the actions of the environmental loads,such as winds,waves,currents,as well as wind turbine aerodynamic loads,the vibration response signals of offshore wind turbine support structures under operation states contain both the harmonic response components induced by blade rotation and structural modal response components induced by the environmental loads.However,the changes of structural dynamic characteristics caused by damage are usually contained in the structural modal responses.Therefore,the modal response components should be first extracted from the structural vibration response signals,and then the modal response is used to accomplish the structural damage identification.Thus,in this paper,the variational modal decomposition(VMD)and probability density function(PDF)are employed to extract the structural modal responses from the vibration response signals under the operational states,the modal responses are used to construct the damage characteristic index,then,the gradient boosting decision tree(GBDT)and support vector machine(SVM)are used to carry out the structural damage identification.The main contents are as follows:(1)In this paper,the research background and significance of structural damage identification are reviewed,and the state-of-the-art of studies on the structural damage identification of offshore wind turbine support structure at home and abroad are introduced.(2)Some basic theories used in this paper,such as VMD,PDF,GBDT and SVM,are introduced,and the selections of their parameters are also given.(3)Modal responses extraction based on VMD and PDF.The numerical model of NREL 5MW offshore wind turbine is constructed by using Bladed software.The acceleration response signals of the wind turbine tower in the shutdown state and operational state are calculated respectively.In view of the characteristics of wind turbine support structure responses with dense frequencies and the harmonic responses with high proportion of energy under the operational state,the response signals are decomposed by VMD to obtain a series of component signals,and the PDF of each component signal is used to select the modal response.(4)Structural damage alarming based on modal response energy.The modal response components are extracted from the original acceleration response signal by using VMD and PDF.Euclidean distances of energy vector of each mode response before and after structural damage are calculated and used as damage alarming index to carry out structural damage alarming.The effects of noise and wind speeds are also considered.(5)Structural damage degree identification based on modal responses.Damage feature vectors are constructed by combing the energies of structural modal response and the energy entropies,and classification algorithm of GBDT and SVM are employed to identify the structural damage degree,separately.The effects of noises and wind speeds are also considered.